dynamically set buffer length

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

Is there a way in Enterprise COBOL to dynamically set the buffer length for a PUT?
Here's what i am trying to do. I want to write a callable subroutine that just does a PUT1.
The subroutine will can be called by many other COBOL programs that will all have different lengths for the message they want to PUT.
I don't want to pick some arbitrary size for the buffer-length. Because as soon as I do someone will exceed it.
I could have them pass in the size of the message they are putting but how does that help?
I will have a PIC in the linkage section that is of some size to hold the message they passed. Can I max out an 01 in the linkage at 16MB and then somehow redefine it to be what they passed in?
I know how to dynamically allocate storage using CEEGTST but still I have to have something of a set PIC defined in that linkage.
Any thoughts?

dbzTHEdinosauer · Posted: Thu Apr 07, 2011 12:37 pm

originator of the message knows the length - it should be passed.

you can put from linkage, so your cobol routine does not need to establish working-storage or dynamic storage -
that is a waste, not only do you double the storage requirements of the task,
there is the move involved.
mqs does a move anyway, why add an extra move.

define your linkage to the max allowable definition.
then you will only have to change when you change compilers.

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

ok. so if I have this in the LINKAGE

01 WS-LINKAGE-DATA.
05 WS-MESSAGE PIC X(100000000).

because the maxmessage length is 100MB and the calling program only passes in a 1000 bytes message through the linkage and then I do a PUT aren't I putting the 1000 bytes plus the rest as spaces??
This seems wasteful too.

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

Plus it doesn't work I get a 2030 if the buffer-length is different than the queue length

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

Sorry that should have read

Plus it doesn't work I get a 2030 if the buffer-length variable is different than the actual length of what you put in the buffer

dbzTHEdinosauer · Posted: Sat Apr 09, 2011 11:19 am

1. I said that the originater (CALLer of your module) kinows the length of the message that was built.
The length of the message and the message should be part of your linkage and the CALLer's USING parms.

2. The mqs parms include

message length
buffer length
address of message/message

parms

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

OK I tried this. The CALLer of the module had this working storage:

01 WS-LINKAGE-DATA.
05 WS-LENGTH PIC 9(08).
05 WS-MESSAGE PIC X(20).

And the module has this linkage

01 WS-LINKAGE-DATA.
05 WS-LENGTH PIC 9(08).
05 WS-MESSAGE PIC X(100000000).

Then my put looks like this
ALL 'MQPUT1' USING PUT-HCONN
OBJECT-DESCRIPTOR
MESSAGE-DESCRIPTOR
PMOPTIONS
PUT-BUFFER-LENGTH
WS-MESSAGE(1:PUT-BUFFER-LENGTH)
PUT-COMPCODE
PUT-REASON.

The problem with this is that if you DISPLAY WS-MESSAGE when you start up the program, what I see in there is a 20 byte message and then a bunch of garbage. It looks like this (Bold text is the actual message):

WS-MESSAGE = PO NUMBER 1234567890
LICENSED MATERIALS - PROPERTY OF IBM5655-DB2(C) COPYRIGHT IBM CO
OVERNMENT USERS RESTRICTED RIGHTS - USE, DUPLICATION OR DISCLOSURE RESTRICTED BY
. DSNAA 11/22/9615.57 0 00 DSNELI 96.327 .} 00 .}
- AM {<j {< \{<;\AM -- N Aq- {bN - A {b 0- - - / & &K K 0
{ K A.K A} 0A K AyK cA K A K A K A K A{K A.K `
AkK A K Am& z A*K Ao 0/@j /q A!K A }} m /q A j
CIB / 51006SQLCA hDSNELI E 00 IGZEBST
N } D# { }* 0 { 0{ N } Dc {> 0 { 0{

However this does work. But I don't know how I feel about the garbage being in there. Seems like a bad idea to have that in there. The only way to get rid of it it to initialize WS-MESSAGE but then you've wiped it out.
Surely I am not the first person to attempt this? There has to be a way that is cleaner

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

Above, one of your mq params was Address. Do you mean pass a POINTER?

dbzTHEdinosauer · Posted: Mon Apr 11, 2011 7:39 pm

1. yes, this has been done more than once.

2. what are you populating PUT-BUFFER-LENGTH with? it is for a put operation, not really used. it is a get operation data-item that tells
mqs when it may overflow with a message. You should be more interested in providing mqs with the actual message length.....

3. what is this: WS-MESSAGE(1:PUT-BUFFER-LENGTH)??
There is no need to provide reference modification. there is a parm for mqs that tells it what the length of the message is.

4. why are you using an unsigned display numeric to pass 'pure numeric' data.

5. why are you worried about the data in the message area beyond the message length? and what is not clean??? you have defined a meg of memory as a potential USING parm. mqs, and everybody else is only worried about the message-length portion of the message.
you want to display a clean message... display ws-message(1:msg-length)

as far as passing pointers or addresses or whatever,
you will have to refer to the MQS documentation to determine what the different parms are and how and with what are they populated.

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

dbzTHEdinosauer · Posted: Mon Apr 11, 2011 8:45 pm

1. been a while since i coded mqs, but there is definitely two distinct parms:
buffer-length
message-length.

2. i have never know ibm to use display numeric variable for a numeric field.
usually they are halfword, or word signed binary:

pic s9(9) comp. or pic s9(4) comp.

3. your phrasing 'seeing garbage get passed into my module.' portends a problem of understanding.

what you see, is the calling modules area, and that you can access it,
means your ss-range compiler options are set to off, which means you can walk on the area of other modules. in development, this is not a good idea. but for purposes of your demo, it does not matter.

in case you do not have it: this is a link to the WebSphere MQ Library

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

Well, I tried using compiler option SSRANGE. It still picked up other areas of the module. I am not sure why.
I appreciate your help. it does work, I don't really think that it picking up other parts of the module is a problem although it bugs me that when you display the message for debugging there is additional stuff in there. That could confusing to someone who didn't know why.

dbzTHEdinosauer · Posted: Tue Apr 12, 2011 2:48 am

you have a linkage pointer (using WS-LINKAGE-DATA)
which addresses 100000000 + 8 bytes of someone else's storage.

you have a length. when you do the display,
DISPLAY WS-LINKAGE-DATA(1:WS-LENGTH + 8)
and you will only display

the length field (a length of 8)
and the message from WS-LINKAGE-DATA for the length of WS-LENGTH

.

or you could have two displays:

Susan Talbot · New User Joined: 17 Dec 2010 Posts: 36 Location: KY

I tried the SSRANGE option and it didn't seem to work although I will ask someone who works on mainframe tech.
The program seems to work fine now and do what I need it to do. Really appreciate the help.
I was worried about other developers wondering what was in the message if they tried to debug but I am sure they will know. I will put a Display in there using ref mod like you show above and it should be ok
Thank you again

dbzTHEdinosauer · Posted: Tue Apr 12, 2011 7:21 pm

always happy to help bluegrass.